Lucas Henrique Soares Guimarães

Serve as a Staff-level individual contributor across health-tech software, contributing to architecture, backend and frontend systems, APIs, cloud services, integrations, and applied AI. Own complex technical work from design through implementation, emphasizing scalable service boundaries, reliability, maintainability, and pragmatic delivery.

Lead architecture and delivery of software and AI capabilities for learning-solutions initiatives, working across backend services, frontend applications, APIs, integrations, and cloud infrastructure. Design and implement AI-enabled workflows and production services using Python/FastAPI, TypeScript/Node.js, React, asynchronous processing, and API-first patterns.

Company ownership and founding engineering role; shape technical architecture and delivery for real-time computer vision and distributed systems. Architected GPU-accelerated computer vision platforms using NVIDIA DeepStream, Triton Inference Server, Apache Kafka, Kubernetes, Python, C++, and GStreamer.

First AI Engineer hire in Brazil for the initiative, helping establish LLM-powered engineering practices and reusable patterns across multiple teams. Architected an AI-driven impact-assessment platform for configuration changes across code, tax, and compliance, reducing manual review cycles by 60%.

Contributed across TypeScript/Node.js backend services and React/Next.js workflows for a dental scanning and case-management platform used by clinicians. Optimized APIs and frontend rendering, improving interaction speed by 35% and reducing page-load time by 1.5 seconds.

Rebuilt core Python/FastAPI services for asynchronous image uploads and background-job orchestration using message queues. Integrated AI-driven product capabilities using OpenAI APIs, improved database performance by 30%, and reduced production incidents by 70%.

Provided tier-3 engineering support for a backup and disaster-recovery platform, resolving 92% of escalated tickets within SLA. Improved triage and root-cause workflows, reducing ticket-resolution time by 30% and repeat incidents by 22%.

Built GPU-accelerated video-inference pipelines with Python, GStreamer, CUDA, and custom plugins for production computer vision workloads. Improved pipeline efficiency by 45%, throughput by 40%, and GPU memory utilization by 20% while preserving production reliability.

Designed and built Windows kernel-mode drivers in C++ to enable integration and adaptation of MBIM cellular modems for Samsung notebook platforms. Worked at the Windows kernel/driver boundary on low-level device integration, reliability, performance, and communications behavior for notebook hardware.

Built a travel recommendation platform end-to-end with Python and React and delivered the MVP 3 weeks early.

Built an eye-tracking application with Python, C++, and OpenCV, achieving 92% accuracy. Released an open-source Node.js/React solution in collaboration with Google engineers to support model optimization and accessibility work.

Completed a Full Stack Boot Camp in 2021.

Began engineering studies in Brazil before continuing later years at the University of Minho in Portugal.

Pursued a five-year integrated master's program in Industrial Electronics and Computer Engineering, advancing through the final year without degree conferral due to interruption during the COVID-19 period.

Completed Windows Kernel Development training in 2022.